Gpt Image2 Ppt
Turns a markdown outline into 16:9 slide PNGs and a .pptx via OpenAI gpt-image-2
Test report
- Verdict
- Works with setup
- Score
- Tested
- Jul 21, 2026
- Environment
- Claude Code 2.x (agent harness)
- Upstream re-checked
- Aug 10, 2026 · 5c2e0f5
Cloned the repo and ran the pipeline for real on a 6-slide investor-pitch deck: wrote slides_plan.md in the documented "## N. [page_type] title" format, md_to_plan.py converted it cleanly (6 slides, page_type parsed), and render_template.py --check correctly detected Keynote 14.0 on this Mac. generate_ppt.py --style styles/clean-tech-blue.md ran the whole chain and stopped at the API call with a clear "缺少 OPENAI_API_KEY" error but still wrote prompts.json, so I could inspect real artifacts: distinct layouts assigned per page (cover-hero-composition / content-structured-grid / comparison-two-zone / data-visual-callouts / closing-summary-contact) and 1.8-2.3k-char prompts carrying exact hex tokens, layout description, content capacity and auto-extracted metrics (84万, 31, 118%, 3.1) — far richer than my baseline python-pptx deck in default Calibri. Two real defects found: any page assigned comparison-two-zone loses ALL body content from the prompt (38 chars of content block vs 417 for a grid page) and a hand-written slide_spec is not injected there either; and LANGUAGE_FONT_RULE in generate_ppt.py hard-codes "all slide text must be Simplified Chinese" with no CLI flag, which neither the English README nor the skill description mentions. Security scan clean (only api.openai.com, base64 is image encode/decode, install script uses $HOME paths, no curl|sh), though it pip-installs into the ambient Python and SKILL.md is 56 KB / 812 lines of always-loaded context.
Scored on four weighted criteria — install, triggering, output vs. baseline, docs. How scoring works
- Installs cleanly 5/5
- Triggers reliably 4/5
- Output vs. baseline 5/10
- Docs & honesty 3/5
What Gpt Image2 Ppt does
Converts a markdown slide outline into slides_plan.json, assigns a page layout from one of 33 bundled style layout banks (or from a rendered .pptx template you supply), builds a per-page image prompt and calls OpenAI gpt-image-2 to render each slide, then packages the PNGs into a 16:9 .pptx. Triggers when a user asks to make a presentation, deck, pitch deck, courseware or slides, or asks to copy the layout of an existing .pptx template. Requires OPENAI_API_KEY and paid gpt-image-2 calls; template-clone and editable modes additionally require PowerPoint, Keynote or LibreOffice locally.
How to install Gpt Image2 Ppt
mkdir -p ~/.claude/skills
git clone --depth 1 https://github.com/JuneYaooo/gpt-image2-ppt-skills.git ~/.claude/skills/gpt-image2-ppt
# The repo's install_as_skill.sh copies to ~/.claude/skills/gpt-image2-ppt-skills (name differs from the slug).
# Needs OPENAI_API_KEY — the script does NOT create .env. Put it in ~/.claude/settings.json:
# "env": { "OPENAI_API_KEY": "sk-...", "GPT_IMAGE_MODEL_NAME": "gpt-image-2" }
# pip install -r ~/.claude/skills/gpt-image2-ppt/requirements.txt
# Only for --template-pptx / --editable: brew install --cask libreoffice (or have Keynote/PowerPoint)
# Verify a render backend exists: python3 scripts/render_template.py --check
Skills live in ~/.claude/skills/ (global) or .claude/skills/
(per-project). Restart Claude Code after installing.
Commands — how to trigger Gpt Image2 Ppt
-
/gpt-image2-pptTurns a markdown outline into 16:9 slide PNGs and a .pptx via OpenAI gpt-image-2
It also activates on plain-language prompts like these:
-
Make an investor pitch deck with this style -
Generate slides mimicking my uploaded PPTX template -
Create a magazine-style presentation on this topic
Frequently asked questions
- Is the Gpt Image2 Ppt skill free?
- Yes. The skill itself is free from JuneYaooo/gpt-image2-ppt-skills. SkillProof publishes the install command and an independent test verdict at no cost.
- Does Gpt Image2 Ppt work with Claude Code?
- We tested it with Claude Code 2.x (agent harness) on Jul 21, 2026. Verdict: Works with setup. Cloned the repo and ran the pipeline for real on a 6-slide investor-pitch deck: wrote slides_plan.md in the documented "## N. [page_type] title" format, md_to_plan.py converted it cleanly (6 slides, page_type parsed), and render_template.py --check correctly detected Keynote 14.0 on this Mac. generate_ppt.py --style styles/clean-tech-blue.md ran the whole chain and stopped at the API call with a clear "缺少 OPENAI_API_KEY" error but still wrote prompts.json, so I could inspect real artifacts: distinct layouts assigned per page (cover-hero-composition / content-structured-grid / comparison-two-zone / data-visual-callouts / closing-summary-contact) and 1.8-2.3k-char prompts carrying exact hex tokens, layout description, content capacity and auto-extracted metrics (84万, 31, 118%, 3.1) — far richer than my baseline python-pptx deck in default Calibri. Two real defects found: any page assigned comparison-two-zone loses ALL body content from the prompt (38 chars of content block vs 417 for a grid page) and a hand-written slide_spec is not injected there either; and LANGUAGE_FONT_RULE in generate_ppt.py hard-codes "all slide text must be Simplified Chinese" with no CLI flag, which neither the English README nor the skill description mentions. Security scan clean (only api.openai.com, base64 is image encode/decode, install script uses $HOME paths, no curl|sh), though it pip-installs into the ambient Python and SKILL.md is 56 KB / 812 lines of always-loaded context.
- What is the Gpt Image2 Ppt SkillProof Score?
- 6.8/10 — installs cleanly 5/5, triggers reliably 4/5, output vs. baseline 5/10, docs & honesty 3/5.
- How do I install Gpt Image2 Ppt?
- Copy the install command from this page, run it in your terminal, and restart Claude Code. Skills live in ~/.claude/skills/ (global) or .claude/skills/ inside a project.
- Can I use Gpt Image2 Ppt with Cursor, Copilot, Gemini CLI, Codex or other AI tools?
- The SKILL.md format is native to Claude (Claude Code, Desktop, claude.ai). The instructions inside adapt to other assistants: Cursor rules, GitHub Copilot instructions, Windsurf rules, Custom GPTs, AGENTS.md for OpenAI Codex, and GEMINI.md for Google Gemini CLI — our conversion guides cover each, and the free converter on the tools page does the wrapping for you.